In middle-age, level of blood pressure (BP) is a strong and independent predictor of both total and cardiovascular disease (CVD) mortality. However, among the elderly, recent short-term observational studies, including our own in the East Boston Established Populations for Epidemiologic Studies in the Elderly (EPESE), have raised the possibility of a J-shaped curve. In our study of short-term mortality, low Bps were positively associated with total and CVD mortality. However, a further analysis confined to the 2079 (68%) of this cohort who were also screened nine years previously for the Hypertension Detection and Follow-Up Program (HDFP) showed the more conventional positive linear relationship between systolic BP, and no association between diastolic BP and mortality. The observations of higher short-term total and CVD mortality at lower BP levels among the elderly could reflect a particular susceptibility to deleterious consequences of drug therapy for high BP. Alternatively, such a finding may be antifactual, due to survival bias or confounding by co-morbid conditions. We plan to study further the association between BP mortality among those who participated in both the HDFP screen and the EPESE study. Linking data from the two separate studies, we will examine change in BP, antihypertensive medication use, and mortality over a 15-year period. With regard to the association between lower BP and increased mortality, we will determine whether the relationship previously observed reflects a fall from normotensive levels as opposed to a consistently low BP. With regard to the upper end of the mortality curve, we will assess the effect of antihypertensive drug treatment on mortality as compared to those with untreated hypertension. This latter question has particular public health significance because questions remain about the risk-to-benefit ratio of antihypertensive drug treatment in this age group. Furthermore, we will also be able to describe changes in BP in this population and medication usage patterns over time and to correlate type of medication and mortality.